2-tier Backup Technology

Download

Introduction

Many companies want to have ability to use backup application in offline mode. Also some customers want to speed up restore process. To satisfy the both groups of customers we introduce a new tier that could replace backup server, it duplicates all server data and uses local file system or network share as storage.

Operations on tiers

All operations on a tier are performed with reconnect procedure with limited attempts (initial implementation will do 10 attempts). A read operation returns result from the first online tier. Any write operation applies to all online tiers, offline and non-synchronized tiers are skipped. A write operation cannot be completed until it succeeds on at least one tier. The last means if execution of a write command is started when all tiers are offline it will wait when one of tiers comes online and the command is applied to it successfully.

Tiers synchronization

Unsynchronized tier detection starts for all returned and new tiers. It is based on revision mechanism. A tier will be marked for synchronization if its revision number less than maximum revision of all tiers. Synchronization of tiers is recursive synchronization of directories and files within corresponding storages. Synchronization is done when application is in idle state. After synchronization tier’s revision is set to the maximum revision explicitly and tier marks as online. Any write operation aborts all in-progress synchronizations.

Second installation detection

Backup client stores revisions of all tiers locally: server revision and n-th tier revisions. If locally stored revision of a tier differs with revision stored within the tier it means there is another installation for this contract. In such case backup application reports error and terminates. Revision of tier is updated before any write operation affected the tier.

2-tier Backup Technology

The main goal of this technology is offline backup availability when FTP server is unavailable. In this case backup will run and send data only to the specified vault location (local or network) using LocalSpeedVault functionality.

As soon as the server is reachable - both databases from server side and vault location will be synchronized and your data will be equal:

Download